I’ve been following the developments in generative AI since ChatGPT was released in November 2022. At first, many of the higher education conversations that I read or heard were focused on a fairly narrow question: will students use AI to cheat?
That question is still important. But it is no longer the only question college leaders need to ask.
AI now affects teaching, learning, advising, assessment, research, student support, workforce preparation, institutional operations, and governance. Artificial intelligence is not simply an academic integrity issue. It is a higher education strategy issue.
The Stanford 2026 AI Index’s education chapter frames the challenge well. It assesses how AI is reshaping education and what that means for teaching, learning, and career readiness. The chapter reports that four out of five U.S. high school and college students now use AI for schoolwork, while policies have not kept pace. Students most commonly use generative AI for research, essay editing, and brainstorming. (Stanford HAI)
That finding should get the attention of every president, provost, dean, faculty senate leader, chief information officer, and trustee. Students are using AI, faculty are experimenting with it, vendors are adding it to existing platforms, and employers are expecting graduates to understand it. The question is not whether AI will enter higher education. The question is whether institutions will proactively shape its use or simply react to it.
What Do We Mean by AI in Higher Education?
Artificial intelligence in higher education refers to the use of AI tools and systems in teaching, learning, research, advising, student support, assessment, administration, and institutional decision-making. Generative AI is the most visible current form because tools like ChatGPT, Claude, Gemini, Copilot, and other systems can generate text, code, images, summaries, lesson plans, rubrics, feedback, and analyses.
The U.S. Department of Education defines AI as “automation based on associations.” Its 2023 report explains that AI moves computing beyond conventional edtech in two important ways: from capturing data to detecting patterns in data, and from providing access to instructional resources to automating decisions about instruction and other educational processes. The Department also warns that those automated decisions create risks of bias, unfairness, and governance failure. (ERIC)
That definition is useful because it explains why AI is different from many earlier education technologies. A learning management system organizes content. A video platform distributes content. A plagiarism checker flags text. AI tools can generate, summarize, recommend, assess, converse, and act.
That is why institutional leadership is vital.
AI Adoption Is Already Ahead of Policy
One of the most common mistakes institutions make is assuming that AI adoption begins when the institution launches an AI initiative. In reality, adoption began long before that.
Students adopt tools because they are convenient. Faculty adopt tools because they save time or help with course preparation. Staff adopt tools because they reduce repetitive work. Vendors adopt AI because it makes their products appear more current. By the time an AI committee is formed, informal adoption may be widespread.
EDUCAUSE’s 2025 AI Landscape Study is useful because it shows that higher education is grappling with AI across strategy and leadership, policies and guidelines, use cases, the higher education workforce, and the institutional digital AI divide. The study’s survey was distributed in November 2024, which means many of the institutional issues it identified were already visible well before the current wave of campus AI activity. (EDUCAUSE)
Tyton Partners’ Time for Class 2025 report shows how quickly personal use is spreading. As of spring 2025, 42 percent of students, 40 percent of administrators, and 30 percent of instructors reported using generative AI weekly or daily. The same report found that only 28 percent of institutions had formal generative AI policies in place, while another 32 percent were still developing them. (4213961.fs1.hubspotusercontent-na1.net)
That gap between use and policy is the uncomfortable reality for many colleges and universities. Students and faculty are not waiting for the perfect institutional framework. They are already experimenting.
As I noted in AI Analysis and AI Reporting Tools Are Getting Way Better, these tools are improving quickly. The fact that an AI tool was unreliable a year ago does not mean it is unreliable today. The fact that a tool is impressive today does not mean it should be used without guardrails. Both statements can be true.
Academic Integrity Is the First Challenge, Not the Only Challenge
Academic integrity remains the issue that brings many faculty and administrators into the AI conversation. That is understandable. Generative AI can write essays, solve problems, produce code, summarize readings, create citations, and polish mediocre work into something that appears competent.
Faculty are concerned. A 2026 report from Elon University and the American Association of Colleges and Universities surveyed 1,057 U.S. faculty members in November 2025. In that non-scientific sample, 95 percent said generative AI would increase student overreliance on AI tools, 90 percent said it would diminish students’ critical thinking skills, and 83 percent said it would decrease student attention spans. The report’s own methodology notes that the sample was non-scientific and not generalizable to all college faculty, but the concerns it documents are consistent with what many campus leaders are hearing. (Elon University)
The mistake would be to treat academic integrity as a detection problem only. Detection tools are imperfect. More importantly, they do not answer the deeper question: what learning is the assignment designed to measure?
If the purpose of an assignment is to assess a student’s unaided writing, then AI-generated drafting may violate the learning goal. If the purpose is to assess a student’s ability to evaluate sources, structure an argument, and revise with feedback, then limited AI use might be allowed with disclosure. If the purpose is to prepare students for a profession where AI-supported writing or analysis is already common, then banning AI entirely may be less realistic than teaching students how to use it well.
Institutions need policies that help faculty distinguish between acceptable, questionable, and unacceptable uses of AI. For example, using AI to brainstorm ideas may be acceptable in one course. Using AI to generate an entire essay and submit it as one’s own work should not be. Using AI to check grammar may be acceptable in one assignment. Using AI to fabricate sources should not be. Using AI to practice coding may be encouraged in one course. Using AI during a closed-book exam may be prohibited.
The key is clarity. Students should not have to guess whether AI is allowed. Faculty should not have to invent entirely new rules in isolation. Syllabi, assignment prompts, rubrics, student handbooks, and academic integrity policies all need to be updated.
I wrote about a recent issue in The Uncomfortable Lesson of “Law Professors Prefer AI Over Peer…”. The uncomfortable lesson is not just that AI can produce polished work. It is that many of our assignments may need to be redesigned if we want to know what students actually understand.
Faculty Need AI Support, Not Just AI Rules
Faculty members are often positioned as the front-line defense against student misuse of AI. That framing is incomplete and, in many cases, unfair.
Faculty are also potential users of AI. They can use AI to create examples, draft quiz questions, develop case studies, generate discussion prompts, translate complex concepts into simpler language, design rubrics, summarize research, support accessibility, and provide faster preliminary feedback. In some fields, AI tools may also be part of the professional practice students need to learn.
Tyton’s Time for Class 2025 report found that generative AI can increase or decrease faculty workload depending on how it is used. Instructors reported workload increases related to monitoring cheating, learning AI tools, and redesigning assessments. At the same time, instructors also reported workload decreases in areas such as developing course content, grading, communicating with students, course-related research, and providing tailored feedback. (4213961.fs1.hubspotusercontent-na1.net)
That finding makes sense to me. AI does not automatically save time. It saves time when users know what they are doing and when the institution supports thoughtful use.
A good faculty AI strategy should include discipline-specific examples, model syllabus language, assignment redesign workshops, privacy guidance, accessibility guidance, and clear expectations for disclosure. It should also include faculty communities of practice. Faculty members are more likely to trust practical examples from colleagues than abstract technology enthusiasm from vendors.
The question for leaders is not whether faculty should use AI. Many already are. The better question is how institutions can help faculty use AI in ways that improve teaching while protecting academic quality.
Student Outcomes Should Be the Measure of AI Strategy
Higher education should not adopt AI because it is fashionable. It should adopt AI where there is a plausible path to better learning, better support, better persistence, better completion, better affordability, or better workforce preparation.
That is why student outcomes should be at the center of any AI strategy.
AI may help students by offering practice problems, explanations, writing feedback, tutoring support, translation assistance, accessibility support, and study planning. It may help advisors by identifying students who need support earlier. It may help faculty by giving them more time for meaningful interaction. It may help institutions improve service responsiveness without adding unnecessary complexity.
But AI can also create false confidence. A student who receives a fluent explanation from a chatbot may believe that he or she understands the concept. A student who uses AI to summarize readings may avoid the difficult work of reading. A student who uses AI to draft every assignment may complete courses while weakening the very skills the course was designed to develop.
This is where student success strategy matters. AI should not be separated from the larger institutional work of improving persistence, completion, and learning. I would connect this directly to What Works Best to Improve Retention and PhD Completion Rates. Student outcomes are rarely improved by one tool. They improve when institutions align expectations, advising, faculty engagement, course design, financial support, and timely intervention.
AI can support that work. It cannot replace it.
Policy Has to Move From “Allowed or Banned” to “Governed”
Early campus AI policies often focused on whether AI was allowed or banned. That was understandable in 2022 and 2023. It is no longer sufficient.
A useful AI policy needs to cover academic use, disclosure, assessment, data privacy, procurement, accessibility, research, intellectual property, cybersecurity, and accountability. It should be clear enough for students to understand and flexible enough for faculty to apply across disciplines.
The U.S. Department of Education’s 2024 AI toolkit is especially relevant for privacy and data governance. It notes that FERPA generally requires educational agencies and institutions to obtain prior written consent before disclosing education records and personally identifiable information, unless an exception applies. It also warns that AI-enabled edtech products may collect lengthy student interactions that could contain sensitive information if safeguards are not in place. (ERIC)
That warning should be taken seriously in higher education. A faculty member pasting student work into an AI system may not realize that student information is being disclosed. A staff member using AI to summarize advising notes may not realize that sensitive information is being processed by a third-party system. A researcher using AI to analyze unpublished data may not realize that intellectual property or human-subjects data is being exposed.
Institutional policy should answer practical questions. What tools are approved? What data may be entered? What data may not be entered? Who reviews vendor contracts? Who evaluates accessibility? Who determines whether an AI tool is appropriate for student-facing use? Who is responsible if the tool produces harmful or biased output?
Those questions cannot be answered one course at a time.
Governance Is Not the Same as an AI Committee
Every institution probably needs an AI committee or working group. But a committee is not governance.
Governance defines decision rights. It identifies who approves tools, who owns risk, who reviews policies, who monitors outcomes, who communicates expectations, and who revises the framework when technology changes.
The National Institute of Standards and Technology’s AI Risk Management Framework offers a useful model. NIST describes the framework as a voluntary tool to help organizations manage risks to individuals, organizations, and society associated with AI. It is designed to improve the ability to incorporate trustworthiness into the design, development, use, and evaluation of AI products, services, and systems. NIST also released a Generative AI Profile in 2024 to help organizations identify risks unique to generative AI. (NIST)
Higher education does not need to turn AI governance into bureaucracy. But it does need structure. A good governance process should include academic affairs, faculty governance, student affairs, information technology, legal counsel, accessibility, institutional research, libraries, teaching and learning centers, procurement, and student representation.
At a minimum, institutions should know which AI tools are in use, what data they collect, what contracts govern them, what risks they create, what benefits they are expected to produce, and how those benefits will be measured.
Governance is not about stopping innovation. It is about preventing unmanaged innovation from becoming institutional risk.
AI Literacy Should Become Part of the Student Experience
AI literacy may become as important as information literacy, digital literacy, and quantitative literacy. Students do not all need to become computer scientists. They do need to understand how AI tools work, where they fail, how to verify their output, how to cite or disclose their use, how to protect private information, and how to use AI ethically within their disciplines.
The Stanford AI Index notes that U.S. four-year computer science enrollment fell 11 percent between 2024 and 2025, while master’s graduates in AI software-related fields rose 17 percent from 2023 to 2024. That combination suggests a more complicated talent picture than a simple “more students are studying technology” narrative. Broad AI literacy and specialized AI expertise are both likely to matter. (Stanford HAI)
AI literacy should be integrated into general education, professional programs, graduate education, continuing education, and workforce partnerships. It should not be limited to computer science departments. Nurses, teachers, accountants, lawyers, engineers, journalists, managers, and public servants will all work in environments shaped by AI.
That is one reason I connect this topic to Agentic AI and How It Will Change Work and Coursera’s 2023 Global Skills Report. If AI changes work, then colleges and universities need to prepare students to work with AI without surrendering judgment to it.
College Leaders Should Ask Better Questions
When I speak with institutional leaders about technology, I often find that the first questions are vendor questions. What platform should we use? What tool should we buy? What are peer institutions doing?
Those are not bad questions. They are just not the best first questions.
For AI in higher education, I would begin with different questions:
What learning outcomes are we trying to protect or improve?
Where are students already using AI?
Where are faculty already using AI?
Which uses are educationally valuable?
Which uses undermine learning?
What data should never be entered into public AI systems?
Which AI tools are already embedded in our existing platforms?
How will AI affect academic integrity, assessment, advising, accessibility, research, and student support?
How will we prepare students for an AI-influenced workforce?
How will we measure whether AI is helping or hurting student outcomes?
Those questions make AI a leadership issue rather than a tools issue.
This is also why boards should be engaged. Trustees do not need to understand every technical detail. They need to understand the implications for academic quality, institutional reputation, privacy, workforce alignment, student learning, and risk management.
A Few Final Thoughts
AI in higher education is not going away. It is already here, and the tools will continue to improve.
The worst response would be to pretend that AI can be kept outside the institution. The second-worst response would be to adopt AI everywhere without strategy, governance, or evidence. The better response is disciplined experimentation.
By disciplined experimentation, I mean a strategy that encourages responsible use, protects student data, supports faculty, redesigns assessment, measures outcomes, and keeps human judgment at the center.
Colleges and universities have adapted to major changes before. They adapted to mass higher education, federal financial aid, online learning, changing demographics, workforce demands, and new accountability expectations. AI is another such moment. It will not eliminate the need for higher education, but it will force institutions to clarify what they value.
If a college’s value proposition is simply access to information, AI is a threat. If its value proposition is guided learning, expert feedback, community, judgment, credentialing, mentoring, research, and preparation for meaningful work, AI can become a powerful tool.
The difference will be leadership.
AI will not answer the most important questions for higher education. It will make those questions harder to avoid.
